Appl Phys A
DOI 10.1007/s00339-012-7081-5
Structural inhomogeneity and piezoelectric enhancement in ZnO
nanobelts
Kasra Momeni Anjana Asthana Abhishek Prasad
Yoke K. Yap Reza Shahbazian-Yassar
Received: 27 July 2011 / Accepted: 5 July 2012
Springer-Verlag 2012
Abstract In this work, piezoelectricity of individual ZnO
nanobelts grown along the [0 1 0] direction is studied using
piezoresponse force microscopy (PFM). It is found that the
effective piezoelectric coefficient of these NBs, d33 eff , is in-
creasing from 2.7 pm/V at 30 kHz to 44 pm/V at 150 kHz.
The results were explained by the Debye model, where
structural inhomogeneity in our NBs was shown to be re-
sponsible for piezoelectric enhancement.
1 Introduction
Wurtzite zinc oxide (ZnO) is structurally noncentral sym-
metric, and thus applicable for piezoelectric devices. As
ZnO has the highest piezoelectric tensor among tetrahedrally-
bonded semiconductors [1], it becomes a promising can-
didate for sensor [2], actuator [3], and energy harvest-
ing [4] applications. In particular, piezoelectricity of ZnO
nanostructures has gained significant attention for nanoscale
piezoelectronic devices. Several techniques including inter-
ferometry [5], scanning tunneling microscopy [6], and scan-
ning probe microscopy (SPM) [79], have been used for
measuring the piezoelectricity of materials. Piezoresponse
K. Momeni R. Shahbazian-Yassar ()
Department of Mechanical Engineering, Michigan Technological
University, Houghton, MI 49931, USA
e-mail: reza@mtu.edu
A. Asthana
Department of Materials Science and Engineering, Michigan
Technological University, Houghton, MI 49931, USA
A. Prasad Y.K. Yap
Department of Physics, Michigan Technological University,
Houghton, MI 49931, USA
force microscopy (PFM), which is a modified atomic force
microscopy (AFM) technique has been introduced recently
for studying piezoelectric and ferroelectric properties of
nanomaterials [10, 11]. A schematic figure of PFM setup
used in our experiments is shown in Fig. 1. In this technique,
an alternative voltage source is connected between the con-
ductive AFM tip and bottom of the sample. The alternat-
ing voltage generates an alternating electric field inside the
piezoelectric sample. The sample responds to the electric
field by mechanical deformation due to inverse piezoelec-
tric effect while AFM tip remains in contact with top of the
sample. Out of plane deformation of the AFM tip will be
measured by monitoring deflection of the AFM cantilever.
In fact, a few studies have been performed on piezoelec-
tric properties of ZnO nanorods and nanobelts (NBs) but
the results are controversial. Scrymgeour et al. [12] showed
that the average value of the d33 piezoelectric coefficient
of solution-grown ZnO nanorods to be around 4.41 pm/V.
Moreover, the measured coefficient was independent of
the frequency of the applied electric field in the range of
150 kHz. In contrast, Zhao et al. [13] have reported that the
piezoelectric coefficient of ZnO NBs grown along the [2 0]
direction at 1400 C was 26.7 pm/V at 30 kHz. This value
reduced as a logarithmic function of frequency to 14.3 pm/V
at 150 kHz. Interestingly, in the above studies, the role of
defects on the electromechanical response was not studied
systematically.
Electromechanical response of piezoelectric materials
depends on their structure and configuration of defects
[11, 14]. Configuration of such defects is in turn affected
by synthesis parameters such as substrate material and an-
nealing temperature [15]. Here, piezoelectricity of ZnO NBs
grown along the [ 01 0] direction with (0001) top and bot-
tom surfaces is reported. This is different from those re-
ported with ZnO NBs grown along the [2 0] direction [13].
 K. Momeni et al.

Fig. 1 Schematic diagram of a

piezoresponse force microscopy
is shown. There is an alternative
voltage source (Vac cos t )
between the conductive AFM
tip and conductive substrate.
Deflection of the cantilever will
be monitored by position
sensitive photo detector (PSPD).
The feedback control on
z-height gives the topography
image while the lock-in
amplifier provides
piezo-response image

We found that the effective piezoelectric coefficients of

[0 1 0] ZnO NBs, d33
eff , increased from 2.7 pm/V at 30 kHz

to 44 pm/V at 150 kHz. Enhancement of d33 eff was explained

in terms of the Debye model, where coupling of elastic and

dielectric effects are responsible for strengthening of piezo-
electric.

2 Experimental procedure

Our ZnO NBs are grown by thermal Chemical Vapor De-

position (CVD) system as reported previously but without
the use of gold catalyst [16]. These NBs are grown along
the [0 1 0] direction as confirmed by transmission electron
microscopy (TEM) [17]. The two major facets of these NBs
are (0001) and (2 0). As-grown ZnO NBs were suspended
in ethanol and a drop of the suspension was put on a Si sub-
strate which was coated with a gold film (30 nm). Then, the
whole sample surface was coated with another 2 nm thick
gold film. The top gold layer acts as an electrode which pro-
vides uniform electric field inside the sample. This method
is based on the integral method of excitation [18], which
monitors the overall piezoelectric response of the sample
rather than its local response. Figure 2 shows the clear con-
trast of a ZnO NB as measured by electric force microscopy
(EFM). This confirmed that the top and bottom gold films
are not short circuited.
The PFM measurements were conducted with a Si can- Fig. 2 (a) Topography and (b) EFM images of a ZnO nanobelt are
shown. The sharp contrast between the NBs and substrate in the EFM
tilever coated with a conductive layer of Platinum-Iridium. image indicates that the NBs are not short circuited
Stiffness of the cantilever was 40 N/m, and radius of cur-
vature of the tip was 10 nm. A contact force was chosen
to make sure that the measurements were in the strong- photodiode response through force-distance curves on a sap-
indentation regime [19]. This is the regime where piezoelec- phire sample [20]. Frequency dependence and background
tric response is dominated by the d33 value of the sample. noise were removed following a background correction pro-
PFM measurements were calibrated by calibration of the cedure described in the literature [21].
Structural inhomogeneity and piezoelectric enhancement in ZnO nanobelts

Fig. 3 (a) Piezoelectric

response of ZnO NB as a
function of applied RMS
voltage at different frequencies
shows linear responses. The
slope of the curves gives the
effective piezoelectric
coefficient of ZnO NBs.
(b) Effective piezoelectric
eff , of the ZnO
coefficient, D33
NBs increases with the
frequency of applied voltage.
Piezoelectric response of PPLN
as a function of frequency is
measured using PFM (red
curve). The piezoelectric
response of PPLN is frequency
independent and is equal to
8.2 pm/V

3 Results and discussion and d33 = 9.93 pm/V [13], d33 eff for the bulk ZnO was cal-

culated to be 4.1 pm/V. The measured d33 eff for ZnO NB

Piezoelectric response of Periodically Poled Lithium Nio-
bate (PPLN) was studied as our standard sample. The piezo-
electric response of PPLN as a function of applied frequency
is shown in Fig. 3b (red curve). It is shown that the piezo-
electric response of PPLN (8.2 pm/V) is independent of
applied frequency and is in good agreement with the theo-
retical value of 8.07 pm/V reported in the literature [22].
The piezoresponse of the NB was measured by applying
an alternating voltage between the tip and the substrate. Fre-
quency of the applied electric field varied between 30 kHz
and 150 kHz, which was far less than the resonance fre-
quency of the cantilever. Deflection of the cantilever as a
function of applied voltage (root mean square, RMS, value
varied from 04 V) is shown in Fig. 3a. The slope of the de-
flection amplitude (Af ) versus applied RMS electric voltage
(Vrms ) gives the effective out-of-plane piezoelectric coeffi-
cient of material, d33 eff [9], where A = d eff V
f 33 rms . As shown
in Fig. 3b (black curve), d33 eff increases with the frequency of
applied voltage (the so-called relaxation response) [23].
Elastic boundary conditions can significantly affect the
measured d33 eff values. If the NB has no constraint to lateral
movement, d33 eff d
33 [24], where d33 is the actual piezo-
electric coefficient. When the NB movement is laterally con-
eff = d 2S13 d , where d and S are com-
strained, d33
33 S11 +S12 31 ij ij
ponents of piezoelectricity and compliance tensors of the
material, respectively. Substituting d31 = 5.1 pm/V [25]
at low frequencies (3040 kHz) was 2.75 pm/V, which
is comparable to that of bulk ZnO [13]. It is interesting to
note that substantially higher piezoelectric coefficients were
obtained at high frequencies. The measured d33 eff at 150 kHz
was 44 pm/V (Fig. 3b, black curve), which was almost ten
times greater than d33eff of bulk material.
The results show that piezoelectric coefficient of ZnO
NBs is strongly frequency dependent, which is in agreement
with Zhao et al. [13] data, but different in the following as-
pects: (1) In the same frequency range, the maximum d33 eff
obtained here is three-time greater than the maximum value
obtained by Zhao et al. [13]. (2) Our d33 eff values increase
with the increase of applied frequency (relaxation behavior).
However, Zhao et al. [13] observed a decrease in piezoelec-
tric response by the increase of applied frequency. Similar
relaxation behavior observed here has also been reported for
single-phase anisotropic ferroelectrics and ferroelectric het-
erostructures [23] and piezoelectric materials with defects
that are both elastic and electric dipoles [26]. The governing
mechanism is referred as the Debye-type relaxation effect.
This mechanism is based on heterogeneity of the material
and coupling between the piezoelectric and dielectric prop-
erties. On the other hand, the retardation behavior [23] has
been assumed to be related to the pinning of defects [13].
In the Debye model, a sample is considered to have de-
fects that are both elastic and electric dipoles [26]. The po-
 K. Momeni et al.

(3b)

where k is the Boltzmann constant, T is the temperature, and

parameters with superscript indicate the value of those
parameters at a very high frequency. Defining the piezoelec-
tric coefficient d as
d
d = d + (4)
1 + i
where d = n/kT . After separation of the real and
imaginary components and defining d = d 0 d , the
piezoelectric coefficient can be defined as follows:
Fig. 4 Bistable potential function with (black curve) and without (red
d0 d (d 0 d )
d = d + = d  () id  ()
curve) external elastic, P , and electric, E, fields. W0 is the height of
i
the barrier and Wen is the energy change due to applying elastic and 1 + 2 2 1 + 2 2
electric fields
(5)
tential energy function can be expressed in terms of bistable Equation (5) has the same form as the well-known Debye
model as shown in Fig. 4. The difference in the electric en- equation [27]. It shows that the total piezoelectric coeffi-
ergy of a dipole and elastic energy of a dipole is responsible cient has two parts, real part and imaginary part. Both real
for piezoelectric enhancement of the material which is for- (d  ) and imaginary (d  ) terms of piezoelectric constants are
mulated as follows. functions of frequency, . Real part of piezoelectric coeffi-
Considering the polarization per volume due to defect cient shows the effective piezoelectric constant, and the sign
(P ) and strain per volume due to defect (x) is defined as of d 0 d determines whether the material shows relax-
follows: ation or retardation behavior. The sign of imaginary part of
P = n (1a) piezoelectric constant determines the rotation direction of
the hysteresis curve, positive sign results in clockwise hys-
x = n (1b) teresis while negative sign results in counter clock-wise hys-
teresis.
where , , and n show electric dipole, elastic dipole, and
In order to verify that our ZnO NBs can be considered
number of equally oriented dipoles per volume, respectively.
under the context of the Debye model, phase imaging us-
The constitutive equations of a piezoelectric material
ing PFM was used to reveal the presence of domains in the
are [26]:
ZnO NBs. The sample preparation steps for phase imaging

 + d E
x = S (2a) were the same as discussed above, except in this case there

was no top gold film on the sample surface so that local
D  +
 = d E (2b) piezoresponse can be revealed. The cantilever spring con-
stant and the applied AC bias were 40 N/m and 500 mV,
where x is the strain vector in Voigt notation, S is the elastic respectively. Phase image of the ZnO NB shows different
compliance tensor,  is the stress vector in Voigt notation,
domains along the axis of the NB (Fig. 5a). These domains
d is the piezoelectric coefficient, E is the electric field vec- can be distinguished as bright contrast (positive phase an-
tor, and is the dielectric permittivity tensor. gles: piezoresponse is phase lead the applied AC phase) and
In presence of an alternating electric field E = E0 eit , dark contrast (negative phase angles: piezoresponse is phase
and P = P0 eit where E0 and P0 are the oscillation ampli- lag the applied AC phase). The distribution of phase oscilla-
tude of piezoelectric fields, respectively, t is time, i is the tion was then characterized along the scanning path (white
imaginary number, and is the frequency of oscillation of line) labeled on Fig. 5a. This scanning path mapped across
the alternating electric source, the dielectric (D) and strain the axis of the NB in Fig. 5b. Blue dotted lines show mini-
(x) tensors are: mum and maximum values of phase along the scanning path
    which are corresponding to the blue crosses in Fig. 5a. As
n2 n
D= + E+ d + shown, the overall phases of these domains are negative (be-
kT (1 + i ) kT (1 + i )
low zero degree) and suggesting for the Debye relaxation
(3a)
    behavior. These domains can be formed due to formation
n n 2
of defects and change in the polarization of material for en-
x = d + E + S +
kT (1 + i ) kT (1 + i ) ergy minimization or due to localized variations in synthesis
Structural inhomogeneity and piezoelectric enhancement in ZnO nanobelts
Fig. 5 (a) Phase image of piezoresponse of the ZnO NB shows do-
mains along the axis of NB, (b) change in the phase of oscillation
along the white line that is shown in (a). Blue doted lines represent
the position of blue crosses in (a)
Fig. 6 (a) Low resolution image of twin boundaries, (b) diffraction pattern taken from a twin boundary region, and (c) high resolution TEM image
of planar defects in ZnO NBs which shows structural inhomogeneity of the material
Fig. 7 Hysteresis curve of the as grown ZnO NB is shown. The upper
curve shows the piezoelectric response of the material when the voltage
increases from 10 V to 10 V (trace), and the bottom curve is the
piezoelectric response of the material when the voltage reduces from
10 V to 10 V (retrace). The hysteresis curve has a clock-wise rotation
direction which is in agreement with the proposed model, in case of
relaxation behavior
conditions during the growth of NBs. Different types of de-
fects may present in the structure of ZnO NBs such as twin
boundaries (Fig. 6a, b), planar defects (Fig. 6c), and vacan-
cies.
Considering materials with relaxation behavior, d =
d(0) d() < 0, the piezoelectric phase angle tan p =
d  ()/d  () is positive, which means the piezoelectric re-
sponse proceeds the excitation. In the case of relaxation
piezoelectric response, the hysteresis curve has a clockwise
rotation angle. The hysteresis response of as grown ZnO
NBs was measured by applying a DC voltage which varies
from 10 to 10 V (Fig. 7).
The hysteresis curve in Fig. 7 shows a clockwise piezo-
electric response. The blue curve shows the piezoelectric re-

sponse of ZnO NBs when the applied electric voltage in-
creased from 10 to 10 V. The red curve shows the piezo-
electric response of ZnO NBs when the voltage has reduced
from 10 to 10 V. Clockwise hysteresis curve indicates that
the piezoelectric response (the imaginary part of piezoelec-
tric coefficient) is ahead of excitation (the real part of piezo-
electric coefficient), which is in agreement with the pro-
posed model for piezoelectric materials with relaxation be-
havior.
Understanding the mechanism governing the frequency
dependence of ZnO NBs would be a major step toward
designing new sensors, actuators, and energy harvesting
devices. Using the governing mechanism, new processing
methods can be introduced for tailoring the piezoelectric
properties of ZnO nanobelts and nanostructures. The results
shown here suggest that the use of ZnO NBs with a bet-
ter control on the defects density and structure may enable
more efficient design of energy harvesting devices for high
frequency sources, such as ultrasonic waves and turbulent
flow.
4 Conclusion
In summary, we found that the piezoelectric properties of
ZnO NBs can have relaxation behavior as a function of
applied frequency. For low frequency measurements, the
piezoelectric coefficient of our NBs was comparable to that
of bulk ZnO. The coefficient was significantly increased
by increasing frequency of the applied electric field, up to
44 pm/V at 150 kHz, sixteen times larger than that measured
at 30 kHz. This is explained by the Debye model, where
structural defects are responsible for formation of elec-
tric and elastic defects. The presence of domains was con-
firmed by PFM phase measurements. These results will form
the basis for better understanding of piezoelectric proper-
ties of one-dimensional materials for their applications in
nanoscale sensors, actuators, and generators.
Acknowledgements The authors would like to thank NSF-CMMI
Grant no. 0926819 and NSF-DMR Grant no. 0820884 for providing
the financial support.
K. Momeni et al.
References
1. A. Dal Corso, M. Posternak, R. Resta, A. Baldereschi, Phys. Rev.
B, Condens. Matter Mater. Phys. 50, 10715 (1994)
2. E. Comini, G. Faglia, G. Sberveglieri, Z. Pan, Z.L. Wang, Appl.
Phys. Lett. 81, 1869 (2002)
3. T. Shibata, K. Unno, E. Makino, Y. Ito, S. Shimada, Sens. Actua-
tors A, Phys. 102, 106 (2002)
4. Z.L. Wang, J. Song, Science 312, 242 (2006)
5. A.L. Kholkin, C. Wutchrich, D.V. Taylor, N. Setter, Rev. Sci. In-
strum. 67, 1935 (1996)
6. R. Winters, M. Reinermann, C. Enss, G. Weiss, S. Hunklinger,
Measurement of the piezoelectricity of films with scanning tun-
neling microscopy. Conf. Proc. 13, 1316 (1995)
7. A.G. Agronin, Y. Rosenwaks, G.I. Rosenman, Nano Lett. 3, 169
(2003)
8. J.A. Christman, J.R.R. Woolcott, A.I. Kingon, R.J. Nemanich,
Appl. Phys. Lett. 73, 3851 (1998)
9. C. Harnagea, A. Pignolet, M. Alexe, D. Hesse, U. Gsele, Appl.
Phys. A, Mater. Sci. Process. 70, 261 (2000)
10. V.V. Shvartsman, A.L. Kholkin, N.A. Pertsev, Appl. Phys. Lett.
81, 3025 (2002)
11. A. Kholkin, I. Bdikin, D. Kiselev, V. Shvartsman, S.H. Kim, J.
Electroceram. 19, 83 (2007)
12. D.A. Scrymgeour, T.L. Sounart, N.C. Simmons, J.W.P. Hsu, J.
Appl. Phys. 101, 014316 (2007)
13. M.-H. Zhao, Z.-L. Wang, S.X. Mao, Nano Lett. 4, 587 (2004)
14. A. Kholkin, Ferroelectrics 221, 219 (1999)
15. L. Schuler, N. Valanoor, P. Miller, I. Guy, R. Reeves, M. Alkaisi,
J. Electron. Mater. 36, 507 (2007)
16. S.L. Mensah, V.K. Kayastha, Y.K. Yap, J. Phys. Chem. C 111,
16092 (2007)
17. A. Asthana, K. Momeni, A. Prasad, Y.K. Yap, R.S. Yassar, Appl.
Phys. Lett. 95, 172106 (2009)
18. Review of Ferroelectric Domain Imaging by Piezoresponse Force
Microscopy
19. S.V. Kalinin, D.A. Bonnell, Phys. Rev. B, Condens. Matter Mater.
Phys. 65, 125408 (2002)
20. M. Alexe, A. Gruverman, Nanoscale Characterisation of Fer-
roelectric Materials: Scanning Probe Microscopy Approach
(Springer, Berlin, 2004)
21. T. Jungk, A. Hoffmann, E. Soergel, Appl. Phys. Lett. 89, 163507
(2006)
22. M. Jazbinek, M. Zgonik, Appl. Phys. B, Lasers Opt. 74, 407
(2002)
23. D. Damjanovic, M.D. Maeder, P.D. Martin, C. Voisard, N. Setter,
J. Appl. Phys. 90, 5708 (2001)
24. J.-F. Li, P. Moses, D. Viehland, Rev. Sci. Instrum. 66, 215 (1995)
25. F. Bernardini, V. Fiorentini, D. Vanderbilt, Phys. Rev. B, Condens.
Matter Mater. Phys. 56, R10024 (1997)
26. G. Bertotti, I.D. Mayergoyz, The Science of Hysteresis (Elsevier,
Amsterdam, 2006)
27. P.J.W. Debye, The Collected Papers of Peter J.W. Debye (Inter-
science, New York, 1954)